Use of SWIRLS Nowcasting System for Quantitative Precipitation
Forecast Using Indian DWR Data
K. Srivastava*, Sharon Lau, H.Y. Yeung, A.M. Kannan*,
S.K.Roy Bhowmik* & H. Singh*
Indian Meteorological Society Symposium ‘TROPMET 2010’ on
“Advances in Weather & Climate Services”,
Kolkata 700020, India, 19-21 May 2010
* India Meteorological Department, Lodi Road, New Delhi-110003
Use of SWIRLS Nowcasting System for quantitative precipitation forecast
using Indian DWR data
Kuldeep Srivastava*, Sharon Lau**, H.Y. Yeung**, A.M. Kannan*,
S.K.Roy Bhowmik*, Hari Singh*
*India Meteorological Department, Lodi Road, New Delhi-110003
** Hong Kong Observatory, Hong Kong
Local severe storms are extreme weather events that last only for a few hours and evolve rapidly.
Very often the mesoscale features associated these local severe storms are not well-captured synoptically.
Forecasters thus have to predict the changing weather situation in the next 0-6 hrs based on latest
observations, an operational process known as “nowcast”. Observational data that are typically suited for
nowcasting includes Doppler Weather Radar (DWR), wind profiler, microwave sounder and satellite
radiance. To assist forecasters in assimilating the weather information and making warning decisions,
various nowcasting systems have been developed by various institutes in recent years. Notable examples
can be found from the list of participating systems in the two forecast demonstration projects organized by
WMO for the Sydney 2000 and Beijing 2008 Olympic, including Auto-Nowcaster (U.S.), BJ-ANC (China-
U.S.), CARDS (Canada), GRAPES-SWIFT (China), MAPLE (Canada), NIMROD (U.K.), NIWOT (U.S.),
STEPS (Australia), SWIRLS (Hong Kong, China), TIFS (Australia), TITAN (U.S.) and WDSS (U.S.). A
common feature of these systems is that they all use rapidly updated radar data, typically once every 6
The nowcasting system SWIRLS (“Short-range Warning of Intense Rainstorms in Localized
Systems”) has been developed by the Hong Kong Observatory (HKO) and was put into operation in Hong
Kong in 1999. The system has since undergone several upgrades, the latest known as “SWIRLS-2” being in
2008 to support the Beijing 2008 Olympic Games. At the invitation of India Meteorological Department
(IMD), SWIRLS-2 is being adapted for use and test at New Delhi in connection with the 2010
Commonwealth Games with assistance from HKO.
SWIRLS-2 ingests a range of observation data including SIGMET/IRIS DWR radar
product, raingauge data, radiosonde data, lightning data to analyze and predict reflectivity, radar-echo
motion, QPE, QPF, as well as track of thunderstorm and its associated severe weather, including cloud-to-
ground lightning, severe squalls and hail, and probability of precipitation. SWIRLS-2 uses a number of
algorithms to derive the storm motion vectors. These include TREC (“Tracking of Radar Echoes by
Correlation”), GTrack (Group tracking of radar echoes, an object-oriented technique for tracking the
movement of a storm as a whole entity) and lately MOVA (“Multi-scale Optical flow by Variational Analysis”).
This latest algorithm uses optical flow, a technique commonly used in motion detection in image processing,
and variational analysis to derive the motion vector field. By cascading through a range of scales, MOVA
can better depict the actual storm motion vector field as compared with TREC and GTrack which does well
in tracking small scales features and storm entity respectively. In this paper the application of TREC and
MOVA to derive the storm motion vector and QPF using Indian DWR data would be demonstrated for a
thunderstorm event over Kolkata.
Keywords: SWIRLS, TREC, GTrack, MOVA, storm motion vector, QPF, Thunderstorm
Convective heavy rainfall event is one of the most disastrous weather phenomena affecting a large
population and of common interest to tropical countries. Accurate forecast of these events are crucial for
early warning of potential hazard to minimize loss of life and property. For the realistic prediction of these
events, there is a need for a very high resolution nowcasting system with sophisticated strategies for
ingesting data of high temporal and spatial density.
For any nowcasting system the most important source of volumetric information on meso-scale in
the current operational observing system is the Doppler Weather Radar (DWR). The installation of four
GEMATRONIC METEOR 1500S model DWRs at Chennai (during the year 2002), Kolkata (2003),
Machilipattanam (2004) and Vishakhapattnam (2006) has heightened the prospects for the operational
implementation of nowcasting system to explicitly predict the evolution of mesoscale phenomena. The DWR
scans with beam width of 1o create 360 beams radials of information per elevation angle. A full volume scan
takes about 15 minutes. This provides high resolution measurement of radial velocity and velocity spectrum
width to ranges of 250 km and of reflectivity to ranges of 300 km.
The Hong Kong Observatory nowcasting system SWIRLS (Short-range Warning of Intense
Rainstorms in Localized Systems) has been in operation since 1999 [Lai & Li 1999]. Its second-generation
version (referred to as SWIRLS-2) has been under development and real-time testing in Hong Kong since
2007. To support the 2008 Beijing Olympic Games, a special version of SWIRLS-2 [Yeung et al. 2009] was
deployed for the Beijing 2008 Forecast Demonstration Project (B08FDP) under the auspices of the World
Weather Research Programme (WWRP) of the World Meteorological Organization (WMO).
The original SWIRLS focused primarily on rainstorm and storm track predictions. The much
enhanced SWIRLS-2 comprises a family of sub-systems, responsible respectively for ingestion of
conventional and remote-sensing observation data, execution of nowcasting algorithms, as well as
generation, dissemination and visualization of products via different channels. It embraces new nowcasting
techniques, namely: (a) blending and combined use of radar-based nowcast and high-resolution NWP
model analysis and forecast; (b) detection and nowcasting of high-impact weather including lightning, severe
squalls and hail based on conceptual models; (c) grid-based, multi-scale storm-tracking method; and (d)
probabilistic representation of nowcast uncertainties arising from storm tracking, growth and decay.
In this study, capabilities of TREC and MOVA techniques of SWIRLS in depicting the storm motion
vector using Indian DWR data is discussed. The motion vector field so derived can then be applied to
forecast the future position of the storm cells or individual reflectivity pixels for QPF.
2.1 Synoptic Observation, Radar Observation & Observed Rainfall
Case selected for this study is the thunderstorm event of 11 May 2009 over W. Bengal. On 11 may
2009 there was cyclonic circulation in lower levels over Bihar & neighbourhood. Trough from this extended
upto extreme south peninsula across Chhattisgarh, Talengana and Rayalaseema. Another cyclonic
circulation hanged over Arunachal Pradesh and adjoining Assam & Meghalaya (Fig.2(a)). These led to
significant moisture incursion at low level over the area. Meanwhile, a trough extended from Arunachal
Pradesh to NW Bay of Bengal in middle troposphere (Fig. 2(b)). At 200 hPa, a significant westerly trough
with jet maxima over the region resulted in strong upper-level divergence (Fig. 2(c)).
(a) 850 hPa (b) 500 hPa (c) 200 hPa
Fig. 2 — Streamline analysis over India, Bay of Bengal and Indochina on 11 May 2009
11:09 UTC 11:39 UTC 14:09 UTC
Fig.3 Radar reflectivity (“MAX” product) as observed by DWR Kolkata on 11 May 2009
Fig. 4 Observed 24-hour rainfall (red scribbles in unit of cm) India ending at 03 UTC on 12 May 2009
On 11 May 2009 Kolkata DWR observed that thunderstorms started developing at 09:39 UTC with
six small meso cells (labeled “A” in Fig. 3(b)) observed in the north-west region about 200 km from Kolkata.
At the same time, another line of echo (labeled “B” in Fig. 3(b)) was observed about 100 km to north of
Kolkata. By 11:09 UTC, the six meso cells moved southeastwards and merged as one large cell about 100
km northwest of Kolkata. Meanwhile the line of echo moved south to about 80 km north of Kolkata. At
11:39 UTC, these cells merged and were seen as one organized east-west band of convections. At 14:09
UTC, the echoes, which continued to head southeastwards to over 100 km southeast of Kolkata, started
dissipating over Bay of Bengal. Corresponding radar images Maximum Reflectivity (Z) are shown in Fig. 3.
The 24-hour accumulated rainfall (between 03UTC 11